Foam Generation and Rheology in a Variety of Model Fractures

Nov 28, 2018 - Gas is used in petroleum reservoirs to displace oil for enhanced oil recovery. The microscopic displacement efficiency of gas is very g...
0 downloads 0 Views 2MB Size
Subscriber access provided by University of Rhode Island | University Libraries

Fossil Fuels

Foam Generation and Rheology in a Variety of Model Fractures Bander AlQuaimi, and William R. Rossen Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.8b02178 • Publication Date (Web): 28 Nov 2018 Downloaded from http://pubs.acs.org on December 6, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Foam Generation and Rheology in a Variety of Model Fractures

B.I. AlQuaimi, W.R. Rossen, Department of Geoscience and Engineering, Delft University of Technology Keywords: Foam flow, In-situ foam generation; Foam in fracture; and Mobility control in fracture

Abstract

Gas is used in petroleum reservoirs to displace oil for enhanced oil recovery. The microscopic displacement efficiency of gas is very good, but at the reservoir scale the process suffers from poor sweep efficiency, especially in naturally fractured reservoirs. Foam can improve the sweep. There have been considerable scientific contributions towards understanding foam flow in nonfractured porous media, with relatively little work on foam flow in fractured porous media. We investigate foam-generation mechanisms in five fully characterized glass models of fractures with different apertures and correlation lengths of the aperture distribution. We also study the rheology of the in-situ-generated foam by varying the superficial velocities of the gas and surfactant solution. We compare the measured pressure gradient against the fracture attributes, aperture and the correlation length of the aperture. We also compare foam texture as a function of position within the fracture as the generated foam propagates through the fracture. Gas mobility was greatly reduced as a result of in-situ foam generation in our model fractures. Foam was generated predominantly by capillary snap-off and lamella division. The measured mobility reduction depends on fracture attributes. Fracture-wall roughness, represented by both the hydraulic aperture and the correlation length of the aperture, plays an important role in foam generation and mobility. Average bubble size increases as the aperture increases, which results in a significant decrease in pressure gradient. Two model fractures show the same two foam-flow regimes central to the understanding of foam in nonfractured porous media: a low-quality regime where pressure gradient is independent of liquid velocity and a high-quality regime where pressure gradient is independent of gas velocity. The mechanisms thought to be behind these two regimes in nonfractured porous media do not apply to these experiments, however.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Introduction Foam is injected to recover the undisplaced oil in petroleum reservoirs. Foam has been applied in the field from as early as the 1960s1. A foam pilot test was conducted in the Snorre Field, starting with laboratory experiments and numerical simulations2-4. Experience and the benefits of steam-foam injection in many field applications are reported in the literature5. Foam is also used in acid diversion for selective stimulation6, 7. Foam was also used for the remediation of an aquifer8. Many petroleum reservoirs have natural fractures caused by earth stresses9. Natural fractures vary in aperture, length, orientation, asperities and wall roughness10-12. Studies have examined foam flow in fractures in the last two decades. Pre-generated foam was injected into sawed rock core samples or blocks to study oil recovery13-15. Pre-generated foam was injected in parallel slits to study sweep and foam rheology16. Studies of pre-generated foam in microfluidic devices as an approximation to fracture flow have been also reported17,

18.

Investigation of foam flow in fracture replicas with

apertures of roughly 30 and 100 µm have been reported19. Foam generation and sweep efficiency in a fractured rock slab with aperture of approximately 100-150 µm was also investigated20. These studies report the pressure gradient across the entire sample, so it is not possible to know how foam texture or pressure evolves as foam propagates through the sample. Moreover, most studies involved the injection of pre-generated foam. In-situ foam generation and propagation in a physicalmodel fracture was reported along with foam texture and pressure gradient versus distance from the injection face in one study21. Fractures can vary in aperture and in the roughness of the fracture wall. It is important to examine how the geometry of the fracture porespace affects foam generation, propagation and mobility. This paper shows the results of in-situ foam generation in five distinct model fractures. The fractures vary in aperture and correlation length of the aperture. The study also addresses foam texture as a function of roughness scale and aperture variation. In addition, it shows the effect of fracture aperture on foam texture and pressure gradient. It is an initial step toward understanding how fracture geometry affects foam properties, which would extend current studies of foam in individual fractures and allow prediction of foam behaviour under more-general.

Materials and Methods

ACS Paragon Plus Environment

Page 2 of 36

Page 3 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Experimental Apparatus and Method Fig. 6 shows the experimental apparatus, the same as that used in a previous study21. Sodium C14-16 alpha olefin sulfonate (Bio-Terge®-AS-40 KSB, Stepan, Voreppe, France), an anionic surfactant with 39 wt.% active component and a critical micelle concentration of 301.0 mg/l, was used in the study. All experiments employed a 1.0-wt.% surfactant solution in demineralized water. The surfactant solution was injected using a Standard Infusion PHD Ultra Syringe Pump (Model-703005, Harvard Apparatus, Holliston, MA, USA). Flow rates are stated to be accurate to within 0.25%, with reproducibility within 0.05% of full scale. This pump is equipped with micro-stepping techniques to further reduce flow pulsation. The pump has a range from 0.0001 µl/hr to 216 ml/min. Nitrogen was injected through a gas mass-flow meter/mass-flow controller (EL-Flow® F-230M-RAD22-K, Bronkhorst High-Tech B.V., Ruurlo, Netherlands) which has a range of 0 - 10 mln/min. The bottom (roughened) glass plate includes four pressure ports, with a distance of 9.0 cm between them, to provide pressure readings across the length of the apparatus. The pressure-difference sensors are signal-conditioned and temperature-compensated. Three different ranges of sensors are used depending on pressure. The sensors (MPXV5004DP, MPXV5010DP and MPXV5050DP, Freescale Semiconductor, Inc., Austin, TX, USA), with ranges of 0 to 4, 0 to 10, and 0 to 50 kPa respectively, have a maximum error of 5.0% from 0°C to 85°C temperature. The sensors were connected to a dataacquisition unit and a computer, where pressure is recorded every second. For monitoring in-situ foam generation and foam texture we used a LEICA MZ 8 Microscope (10445538 1.0X, Leica Microsystems B.V., Amsterdam, Netherlands). The Microscope is connected to DRS’s lightning RDTTM camera, consisting of a small camera head, detachable cable and custom frame-grabber board. The lightning RDTTM is an ultrafast, high-resolution camera that captures 1280 x 1024-resolution images at 500 full frames per second (fps). A higher fps of 16,000 can be achieved at reduced resolution for recording extremely rapid events. MiDAS 2.0 camera-control software (Xcitex Inc., Woburn, MA, USA) was used to process the images/videos in real time during recording. A compact backlight (model CVI STAR-BL-110/110-WH-24V; Stemmer® Imaging B.V.) provided constant and even illumination. Uniform light is needed to produce noise-free images. Three sets of experiments were carried out using this setup, after measuring experimentally the hydraulic aperture of each fracture: 1. In-situ Foam Generation: The fracture was first vacuum-saturated with water (no surfactant), followed by co-injection of gas and surfactant solution. The foam-generation mechanisms within each of the fractures were observed and categorized.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2. Foam Propagation: Once the foam had been generated, its behavior and evolution as it propagated through the model was investigated. 3. Foam-Quality Scan: After foam flow had been established throughout the fracture, the pressure gradient across the four sections was recorded until a stable signal was observed. The variation in the pressure gradient with foam quality, holding total superficial velocity ut constant, could then be recorded.

Figure 1. Schematic of the experimental setup. The injection and production lines are fitted from the bottom of the fracture plate, but are drawn from the top here to avoid clutter in the diagram. The bottom right shows the modelfracture layout with injection and pressure ports.

Model Fractures Model fractures made of glass plates have previously been used to study foam and two-phase flow in fractures16, 21-29,45. Glass-model fractures provide the ability to observe the flow and investigate the mechanisms of foam generation. More importantly, they allow one to systematically vary roughness scales (magnitude of aperture, aperture variation and the length scale over which the aperture varies) and investigate the effect of these on foam generation, stability and mobility. Our goal is to cover a wide range of apertures and different fracture geometries encountered in fractured reservoirs. Figs. 2 to 6 show the fracture-wall surface topography of our model fractures.

ACS Paragon Plus Environment

Page 4 of 36

Page 5 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The model fractures used here consist of a roughened plate to represent the fracture roughness and a top plate that is smooth, to allow direct observation of the flow. One model fracture (Sample 2) has a 40 X 10 cm plate with regular patterns in its roughness. The remaining four model fractures have 43 X 10 cm plates with significant differences between them in their roughness scales. The roughened plate is 4 mm thick and was strengthened by attaching a 15 mm-thick plate of glass to the back using DELO® Photobond® glue (DELO, Windach, Germany). The thickness of the top glass plate was also 15 mm. The required thickness of the glass plates was estimated using solid-mechanics calculations to prevent any glass deflection during the flow. The glass deflection was also checked using a Probe Indicator (2 µm resolution) during the experiment. In all the model fractures the roughened glass plates include three inlet ports that allow a separate co-injection of gas and liquid. These inlet ports are equally spaced and connected to an 8.0 X 2.0 X 0.04 cm entry trough milled into the roughened plate (Fig. 6, bottom right). The middle inlet port was used for liquid injection and the other two inlet ports for gas injection. Sample 2 has a single port for outflow without a milled outlet trough29. The milled outlet trough in the other four samples eliminates radial converging flow to the single outlet port that we observed in Sample 2. For Samples 1, 3, 4 and 5, the roughened glass plates include four pressure ports spaced over a length of 39 cm and an 8.0 X 2.0 X 0.04 cm milled outlet region. The fourth pressure port is located 2 cm upstream of the outlet trough. Thus behavior in the fourth section, between taps 3 and 4, is thus relatively isolated from the capillary end effect at the edge of the fracture. The gap between the top plate and the rough surface represents the fracture aperture. The two glass plates are glued together at the edges using Araldite® 2014, an epoxy adhesive that has a tensile strength of 26 MPa at 23°C. The fracture is mounted in a frame that can slide 50 cm in X and Y directions to allow for microscopic observation of the flow in the whole 43 X 10 cm fracture.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Sample 1: 3D surface topography. The patch shown is 4 x 4 cm.

Figure 4. Sample 3: 3D surface topography. The patch shown is 4 x 4 cm.

Figure 3. Sample 2: 3D surface topography. The patch shown is 1 x 1 cm.

Figure 5. Sample 4: 3D surface topography. The patch shown is 4 x 4 cm.

Figure 6. Sample 5: 3D surface topography. The patch shown is 4 x 4 cm.

A fracture can be considered a two-dimensional network of pore bodies (maxima in aperture) connected by throats (saddle points between pore bodies)30-33. To characterize the two-dimensional

ACS Paragon Plus Environment

Page 6 of 36

Page 7 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

network, a 4 X 4 cm patch of each roughened glass sample was profiled to quantify the spatial and vertical variations in height. Images and statistics of the pore throats and pore bodies were reported previuosly34. We identify the characteristic pore-throat aperture (dt) as that at the percolation threshold, a characteristic pore-body aperture (db) that is the average pore-body aperture, and a characteristic pore length (Lp) that is the average pore-body length of the 2D network in the flow direction. A separate measure would be the correlation length of aperture. Table 1 shows that these two measures correlate well with each other. The hydraulic aperture dh was measured experimentally by injecting water and obtaining the relationship between flow rate and pressure drop35. We estimated pore-throat width wt (Table 1) by estimating the average pore throat on the percolation path and determining the width of that throat at that aperture. Table 1 summarizes the fracture-aperture data for all five fractures. The model-fracture topography in each case, with the conceptual 2D network superimposed on the images, is given in Appendix A. Samples 4a, 4b, 5a and 5b were fabricated to investigate the effect of the hydraulic aperture (dH) at a fixed Lp. The model fractures were fabricated from glass plates similar to Samples 4 and 5, but with spacers with known thickness (and dimensions 0.5 X 0.5 cm) distributed uniformly, mainly at the fracture perimeter. Additionally, four spacers, one in the center of each section, were placed to prevent deformation of the sample. The spacers occupy only 0.01% of the total area available for flow. We measured the hydraulic aperture for each model after fabrication. Table 1 Model fracture aperture and roughness data (all measurements are in µm)

Pore-throat width, wt

Hydraulic aperture (experimentally determined), dH

Porethroat aperture, dt

Porebody aperture, db

Pore Length, Lp

Correlation length, Lcor

Sample 1

670

818

1128

2661

2754

1550

Sample 2

66

68

138

819

795

410

Sample 3

330

443

853

5156

4800

1650

Sample 4

51

100

210

4415

5100

1130

72

121

231

4415

5100

1130

204

253

363

4415

5100

1130

115

131

211

2421

2240

460

Sample 4a Sample 4b Sample 5

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Sample 5a Sample 5b

Page 8 of 36

145

161

241

2421

2240

630

170

186

265

2421

2240

630

Results In-situ Foam Generation We observed foam generation in our five model fractures, with corresponding mobility reduction of the gas. Foam was generated in situ mainly by snap-off and lamella division. In Samples 2, 3 and 4, both lamella division and repeated snap-off occurred. The throats in these samples are wide in the plane of the fracture (Appendix A) but narrow in aperture dt (Table 1); this slit-shaped geometry favours snap-off36, 37,45 (Appendix B). In Sample 2 snap-off created bubbles that are much smaller than the pores21. Lamella division was observed at high gas fractional flow (fg). Fig. 7 shows snap-off events in Sample 2 at fg = 0.37 and total superficial velocity (ut) of 0.0021 m/s, in all images the flow is from right to lift. Fig. 8 shows lamella division at fg = 0.87 and ut = 0.0049 m/s. In these and similar images to follow, the white area represents water, which occupies the peaks in the topography of Figs. 1-5 (i.e., locations of narrowest aperture) and some pore throats (saddle points between peaks; see Appendix A). Gas occupies the pore bodies (i.e., locations of widest aperture, or valleys in Figs. 1-5). Lamellae appear as white lines in Fig. 7 and following. We observe foam generation by lamella division when a lamella leading a large bubble divides as it encounters a split in the flow path. We did not observe lamella division at fg lower than 0.76 in Sample 2, probably because bubbles were too small to divide. In 3D pore networks, Inter-bubble diffusion can rapidly eliminate bubbles much smaller than pores. In our experiments, diffusion does not have time eliminate these bubbles because bubble residence time in our model is relatively short, approximately 2.7 min21. A similar observation of bubbles smaller than pores was reported in another study of foam flow in fractures20. In Samples 3 and 4 we also observed snap-off; see Figs. 9 and 11, respectively. Samples 3 and 4 differ greatly in their hydraulic apertures and correlation length for aperture (Table 1). This led to significant differences in foam texture (cf. Figs. 9 and 11), foam texture as a function of position (discussed in the next section), and pressure response. Moreover, in Samples 3 and 4 lamella division occurred at flow conditions that were similar to those of snap-off in the same samples. Figs. 10 and 12 show lamella division in Samples 3 and 4, respectively.

ACS Paragon Plus Environment

Page 9 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

In Samples 1 and 5, foam was generated primarily by lamella division (cf. Figs. 13 and 14). In Sample 1 the throats are deeper than in the other samples, and thus less slit-like. In Sample 5 the throats are slightly deeper than in Sample 4, for instance, but not nearly as wide in the plane of the fracture (Figs. 4 and 5, Appendix A): again, less slit-like and less favorable to snap-off. The large aperture of Sample 1 has a significant impact on foam texture and pressure gradient, as discussed below. These results show that the foam-generation mechanism is a function of aperture, fracture-wall geometry, gas fractional flow and total superficial velocity. Surfactant-solution type and concentration are also important but they were the same for all experiments.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 7. Sample 2: foam generation by snap-off; image size (0.75 X 0.43 cm). fg = 0.37 and ut = 0.0021 m/s. Black is gas and white is water. Area of interest is highlighted in red.

Figure 9. Sample 3: foam generation by snap-off (see arrow and box); image size (2.6X2.1 cm). fg = 0.45 and ut = 0.0013 m/s. Black is gas and white is water.

Figure 11. Sample 4: foam generation by snap-off (see arrow and box); image size (1.1X0.9 cm). fg = 0.68 and ut = 0.0032 m/s. Black is gas and white is water.

Figure 8. Sample 2: foam generation by lamella division; image size (0.21X0.2cm). fg = 0.87 and ut = 0.0049 m/s. Black is gas and white is water. The divided bubble is highlighted in red.

Figure 10. Sample 3: foam generation by lamella division (see arrow and box); image size (2.6X2.1 cm). fg = 0.45 and ut = 0.0025 m/s. Black is gas and white is water.

Figure 12. Sample 4: foam generation by lamella division (see arrow and box); image size (0.72X0.66 cm). fg = 0.45 and ut = 0.0016 m/s. Black is gas and white is water.

ACS Paragon Plus Environment

Page 10 of 36

Page 11 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 13. Sample 1: foam generation by lamella division (see arrow and box); image size (2.1X1.8 cm). fg = 0.60 and ut = 0.0025 m/s. Black is gas and white is water.

Figure 14. Sample 5: foam generation by lamella division (see arrow and box); image size (1.2X0.96 cm). fg = 0.70 and ut = 0.0007m/s. Black is gas and white is water.

Foam Propagation We monitored the texture of the foam across the fracture at steady-state flow conditions, using images captured at different distances from the injection port. Sample 1 Sample 1 has the widest hydraulic aperture dH (Table 1) and many asperities. We analyzed foam texture for fg = 0.45 and ut = 0.0012 m/s. The analysis shows that gas enters the model and propagates about 6 to 10 cm as a continuous phase (Fig. 15, image 1). This is evident in the average bubble size and the number of bubbles per unit area. The continuous gas phase starts to break up into relatively smaller gas bubbles by lamella division as discussed previously. We did not observe significant changes in foam texture in sections 2 and 3 (Fig. 15, images 2 and 3). However, in the last section the bubble size became somewhat smaller than the average size of the pore, 7.3 mm2, determined from the 2D network analysis (Fig. 15, image 4). Table 2 presents the statistics from the image analysis for Sample 1. We believe that the foam has not reached a final local-equilibrium state in this case. The wide aperture strongly influences the entrance region, lengthening it considerably. This effect was also clear in the pressure response as discussed in the next section.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 36

Figure 15. Sample 1: Foam texture vs distance at fg = 0.45 and ut = 0.0012 m/s. Image size is 2.5 X 1.7 cm; black is gas and white is water. The images were captured once the pressure gradient had stabilized. Gas is initially continuous and bubbles are generated by lamella division as gas propagates through the fracture. Bubble size becomes smaller than the pore-body size toward the last section of the model fracture. The number of bubbles per unit area significantly increased in section 4. Table 2 Sample 1: image-analysis statistics. fg = 0.45 and ut = 0.0012 m/s. Section

1

2

3

4

Distance from inlet, mm

60

150

230

360

58.06

14.23

17.66

4.92

107.6

20.63

21.22

6.19

5

21

17

55

Average bubble size, mm2 Bubble size, std. dev. ,

mm2

Number of bubbles per unit area

Sample 2 Sample 2 has a regular pattern in its roughness, with much smaller dH and Lp than Sample 1. Foam was generated mainly by snap-off and lamella division. In a manner similar to Sample 1, we captured images at different distances from the injection point. In this model fracture, unlike the others, there was converging flow toward a single outlet port, so the last section is not included in the foamtexture analysis28. This analysis was performed at fg = 0.37 and ut = 0.0021 m/s. The foam gets finer as it propagates through the facture, due to snap-off. The average bubble size decreases and the number of bubbles per unit area in section 3 is almost double that in section 1 (Fig. 16 and Table 3). The two tests were not at identical fg and ut, but the fact that in Sample 2 bubble size is so much smaller than in Sample 1 suggests that both dH and Lp play a role in foam texture. By section 3 of Sample 2 the average bubble size was much smaller than the pore body of the sample, which is 0.50 mm2.

ACS Paragon Plus Environment

Page 13 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 16. Sample 2: Foam texture vs distance at fg = 0.37, ut = 0.0021 m/s; black is gas and white is water. The images are captured during stabilized pressure gradient. Image size 0.8X0.77 cm. Foam-texture analysis shows that the average bubble size decreases and the number of bubbles in section 3 is almost double that in section 1. Table 3 Sample 2: image-analysis statistics. fg = 0.37, ut = 0.0021 m/s. Section

1

2

3

Distance from inlet, mm

20

120

270

Average bubble size, mm2

0.250

0.138

0.081

Bubble size, std. dev. , mm2

0.205

0.125

0.056

165

217

303

Number of bubbles per unit area

Sample 3 Sample 3 has the second largest dH and the largest Lp (Table 1), and foam was generated by both snap-off and lamella division as discussed earlier. Foam-texture analysis was performed at fg = 0.60 and ut = 0.0013 m/s and shows that the average bubble size decreases and the number of bubbles in section 4 is 12 times greater than in section 1 (Fig. 17). The average pore-body size of this sample is 32.9 mm2, which is significantly larger than the average bubble size of 4.47 mm2 observed towards the end of the fracture (Table 4). Similarly to Sample 1, we observed large gas bubbles near the entrance, and only towards the last section did the foam bubbles become finer.

Figure 17. Sample 3: Foam texture vs distance at fg = 0.60, ut = 0.0013 m/s; black is gas and white is water. The images are captured during stabilized pressure gradient. Image size 1.7X1.1 cm. The number of bubbles is 12 times greater in section 4 than in section 1.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 36

Table 4 Sample 3: image-analysis statistics. fg = 0.60, ut = 0.0013 m/s. Section

1

2

3

4

Distance from inlet, mm

60

150

230

360

Average bubble size, mm2

34.24

15.73

12.99

4.47

Bubble size, std. dev. , mm2

19.55

18.72

6.93

5.63

2

5

10

24

Number of bubbles per unit area

Sample 4 Sample 4 is characterized by a small dH and a large Lp. Foam was generated by both snap-off and lamella division in this sample. Foam-texture analysis was performed at fg = 0.70 and ut = 0.0016 m/s. This analysis shows that a considerable number of lamellae have been created in section 2, as compared to Samples 1 and 3, where the dH were much larger, 670 and 330 µm respectively (Fig. 18). Foam propagates through the fracture and is refined as it flows downstream. The average pore-body size in this sample is 13.2 mm2, compared to the average bubble size of 0.14 mm2 observed towards the end of the fracture (Table 5). The small dH in this sample has influenced the bubble size greatly, making significant number of bubble within a short distance of fluid entry.

Figure 18. Sample 4: Foam texture vs distance at fg = 0.70, ut = 0.0016 m/s. Black is gas and white is water. The images were captured during stabilized pressure gradient. Image size 1.4X1.0 cm. An image of section 1 was not available for the analysis. Among our samples, Sample 4 has the smallest dH, 51 µm. A considerably greater number of lamellae have been created in section 2 as compared to samples 1 and 3, where dH = 670 and 330 µm, respectively. Table 5 Sample 4: image-analysis statistics. fg = 0.70, ut = 0.0016 m/s. Section

1

2

3

4

Distance from inlet, mm

60

150

230

360

Average bubble size, mm2

NA

0.36

0.26

0.14

NA

0.47

0.40

0.16

NA

207

216

479

Bubble size, std. dev. ,

mm2

Number of bubbles per unit area

ACS Paragon Plus Environment

Page 15 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

In addition to the original Sample 4, with dH = 51 µm, Samples 4a and 4b have dH = 72 and 207 µm, respectively. We made a comparison of foam texture at the same distance from the injection port once a stable pressure gradient was observed in each case. This test was conducted at fixed fg = 0.45 and ut = 0.0032 m/s. Coarser-textured foam is evident as dH increases (Fig. 19), and fewer snap-off events are observed at dH of 207 µm. (The increase in bubble volume is greater than the increase in bubble area as aperture increases (Table 8). The average bubble size increases with increasing dH (Table 6).

Figure 19. Samples 4, 4a, 4b: foam texture versus dH at fg = 0.45 and ut = 0.0032 m/s. Black is gas and white is water. The images were captured during stabilized pressure gradient. The image size is 1.7X1.5 cm. The correlation length of roughness and Lp are the same in all three fractures. Images are captured 36 cm from inlet. The analysis shows that for fixed Lp the average bubble size increases with increasing dH. Table 6 Samples 4, 4a, 4b: effect of hydraulic aperture on foam texture at fixed Lp. Parameter

Sample 4

Sample 4a

Sample 4b

Hydraulic aperture d , µm

51

72

207

Average bubble size, mm2

0.097

0.148

1.37

Bubble size, std. dev. , mm2

0.114

0.133

1.32

972

750

78

H

Number of bubbles per unit area

Sample 5 Foam was generated solely by lamella division in Sample 5. The foam-texture analysis was performed at fg = 0.46 and ut = 0.0007 m/s. Initially, the gas forms a continuous phase, and foam bubbles are created as it propagates downstream. The average pore-body size of this sample is 4.00 mm2, compared to the average bubble size of 0.53 mm2 observed towards the end of the fracture (Table 7). Foam is generated by a similar mechanism in both Samples 1 and 5. The two samples have roughly the same Lp; however, the foam texture is different in the two samples due to the difference in apertures (Fig. 20).

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 36

Figure 20. Sample 5: Foam texture vs distance at fg = 0.46, ut = 0.0007 m/s; black is gas and white is water. The images are captured during stabilized pressure gradient. Image size 1.6X1.6 cm. Initially the gas forms a continuous phase, then bubbles are created by lamella division. Table 7 Sample 5: image-analysis statistics. fg = 0.46, ut = 0.0007 m/s. Section

1

2

3

4

Distance from inlet, mm

60

150

230

360

Average bubble size, mm2

2.48

0.66

0.60

0.53

Bubble size, std. dev. , mm2

7.84

0.57

0.48

0.36

37

160

176

194

Number of bubbles per unit area

Samples 5a and 5b have dH = 145 and 170 µm, respectively. Tests were conducted at fixed fg= 0.45 and ut = 0.0022 m/s (Fig. 21). The image analysis reveals a similar behavior to Sample 4, with coarsertextured foam observed as dH increases. The average bubble size increases, and the number of bubbles decreases, as dH increases (Table 8).

Figure 21. Samples 5, 5a, 5b: Foam texture versus dH at fg = 0.45 and ut = 0.0022 m/s. Black is gas and white is water. The images are captured during stabilized pressure gradient. The image size is 1.1X0.86 cm. The roughness scale, or Lp, is the same for all three fractures. Images are captured 36 cm from the inlet. The analysis shows that for a fixed Lp the average bubble size increases as dH increases.

ACS Paragon Plus Environment

Page 17 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Table 8 Samples 5, 5a, 5b: effect of hydraulic aperture on foam texture at fixed Lp Parameter

Sample 5

Sample 5a

Sample 5b

Hydraulic aperture d , µm

115

145

170

Average bubble size, mm2

0.468

0.74

0.943

Bubble size, std. dev. , mm2

0.343

0.438

1.02

120

55

54

H

Number of bubbles per unit area

Comparison of Samples These experiments demonstrate the effect of dH and Lp on foam texture. In all the samples, foam becomes finer as it propagates through the fracture. We cannot confirm that foam has reached the final local equilibrium state by the time it reaches the outlet in these experiments. Fine-textured foam was observed in the fractures with the smallest apertures and course-textured foam in the fractures with the largest apertures. Samples with approximately similar apertures (Samples 2 and 4) and different Lp show two distinctly different textures: smaller bubbles in the fracture with smaller pores, though the bubbles are smaller than the pores in both cases. Foam occupies the pore bodies differently, based on the shape of the pore bodies.

Foam-Quality Scans Foam-quality scans were carried out on these model fractures, by holding ut constant and varying fg. The surfactant solution and nitrogen were co-injected into the initially water-saturated fracture, and the pressure gradient across the four sections was recorded until stabilization of pressure gradient was achieved. Significant pressure oscillations were observed in these tests, and larger oscillations were evident at high fg. These oscillations reduce the time-average foam apparent viscosity. In nonfractured porous media the foam behavior at high quality is believed to reflect the destruction of foam at the limiting capillary pressure38-40. We did not observe significant foam coalescence in any of our samples at any tested foam qualities. In our experiments oscillations in pressure gradient reflect fluctuations in foam generation28. We selected the fourth section of each sample, except for Sample 2, as the basis for our analysis of the pressure behavior. In Sample 2, we used the third section, due to the converging flow toward the outlet port in the fourth section. We averaged the pressure gradient over the period of stabilization for each foam quality. The injected gas volume was corrected to the pressure at the middle of the fracture.

ACS Paragon Plus Environment

Energy & Fuels

We tested foam mobility as a function of foam quality fg for four total superficial velocities ut, for three of the model fractures (Samples 2, 4 and 5). Foam quality fg was varied in a random sequence, to avoid misinterpreting the possible effects of hysteresis that might occur in the case of sequential increase or decrease in fg. For Sample 1, with dH = 670 µm, foam was observed only towards the outlet of the model. The recorded pressure gradient p was only a few mbar/m (a few hundred Pa/m), with large oscillations (Fig. 22). Although we tested flow at different values of ut and fg, we were not able to obtain a meaningful foam-quality scan on this sample due to large oscillation at very low pressure gradient. Therefore the uncertainty and variability in p was too great for meaningful analysis, especially at higher fg. Similarly, for Sample 3, with the second largest dH of 330 µm, we did not obtain a foam-quality scan. The recorded pressure gradient was an average of 34.56 mbar/m with significant fluctuations (Fig. 23), even at low fg. It was harder to create foam, reduce gas mobility and increase p significantly with wider apertures.

section 4 70

Section 4

70

60

Pressure gradient, mbar/m

Pressure gradient, mbar/m

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 36

50 40 30 20 10 0 -10 -20 0

1

2

3

4

5

60 50 40 30 20 10 0 -10 0

Fracture Pore Volume Injected

Figure 22. Sample 1: (dH = 670 µm); section 4 pressure gradient at ut = 0.0012 m/s and fg = 0.45. No foamquality scan could be carried out due to small magnitude of pressure gradient and large oscillations.

2

4

6

8

Fracture Pore Volume Injected

Figure 23. Sample 3: (dH = 330 µm); section 4 pressure gradient at ut = 0.0013 m/s and fg = 0.45. No foam-quality scan could be carried out due to small magnitude of pressure gradient and large oscillations.

Foam-quality scans were successfully carried out for Samples 2, 4 and 5. In general, as the velocity increases, the pressure gradient increases; however, the increase is not proportional to ut. Figs. 24, 25 and 26 show the foam-quality scans of Samples 2, 4 and 5, respectively. The effect of Lp or the correlation length on the pressure gradient is made clear by comparing Samples 2 and 4, which have the similar hydraulic aperture dH (Table 1). The overall pressure gradient for Sample 2 is greater than that for Sample 4. We believe this is due to the fact that there is a throat which both contributes to foam generation and restricts bubble flow every 800 µm in Sample 2 (Table 1). The throat apertures are somewhat greater in Sample 4, but the pores are also five times longer.

ACS Paragon Plus Environment

Page 19 of 36

Figs. 24 to 26 indicate the range of shear-thinning behaviour in these three samples. For Sample 2, at injected gas fraction fg = 0.25, pressure gradient p increases only about 14% upon an increase in total superficial velocity by a factor of 5: in effect, a power-law exponent n less than 0.1 For fg = 0.75, n is about 0.43. For Sample 2, at fg = 0.3, n ~ 0.3, and at fg = 0.7, n ~ 0.85, nearly Newtonian. For Sample 5, at fg = 0.3, n ~ 0.26, and at fg = 0.7, n ~ 0.83.

Central to the understanding of flow in nonfractured porous media is the existence of two distinct foam-flow regimes, corresponding to high foam quality and low foam quality40, 41. The pressure gradient is independent of liquid velocity in the low-quality regime and independent of gas velocity in the high-quality regime. In Sample 2 these two regimes were observed. Fig. 27 shows the pressuregradient contours for Sample 2. Pressure-gradient data for Samples 4 and 5 are shown in Figs. 28 and 29, respectively. The same two foam-flow regimes were observed in Sample 5. For Sample 4, all of the data would correspond to a transition region between the high- and low-quality regimes. The transition between regimes is sensitive to both the nature of the porous medium and the ability of the surfactant to stabilize foam40. Given the absence of evidence of either flow regime in so wide a scan of foam quality (see Fig. 28), it may well be that the two regimes do not apply to this foam in this fracture.

ut = 0.0049 3500

ut = 0.0030

ut = 0.0021

ut = 0.0077 m/s ut = 0.0032 m/s

ut = 0.0010 1400

3000

Pressure gradient,mbar/m

Pressure gradient ,mbar/m

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

2500 2000 1500 1000 500

ut = 0.0047 m/s ut = 0.0016 m/s

1200 1000 800 600 400 200 0

0 0.00

0.20

0.40

fg

0.60

0.80

1.00

0

0.2

0.4

fg

0.6

0.8

1

Figure 24. Sample 2: foam-quality scans at different total Figure 25. Sample 4: foam-quality scans at different total superficial velocities ut (m/s). The error bars in the data superficial velocities ut (m/s). The error bars in the data reflect oscillations in pressure gradient. reflect oscillations in pressure gradient.

We do not know the reason for this difference. Sample 4 has narrower aperture than Sample 5 and similar aperture to Sample 2 (Table 1). It is possible that foam has not reached local equilibrium with the fracture, since texture is still rapidly changing in the fourth section (Fig. 18, Table 5).

ACS Paragon Plus Environment

Energy & Fuels

ut = 0.0036 m/s

ut = 0.0022 m/s

ut = 0.0015 m/s

ut = 0.0007 m/s

800 Pressure gradient, mbar/m

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 36

700 600 500 400 300 200 100 0 0

0.2

0.4

0.6

0.8

1

fg

Figure 26. Sample 5: foam-quality scans at different total superficial ut velocities (m/s). The error bars in the data reflect oscillations in pressure gradient.

Figure 27. Sample 2: pressure-gradient (mbar/m) as a function of superficial velocities of gas and liquid.

ACS Paragon Plus Environment

Page 21 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 28. Sample 4: pressure-gradient (mbar/m) as a function of superficial velocities of gas and liquid.

ACS Paragon Plus Environment Figure 29. Sample 5: pressure-gradient (mbar/m) as a function of superficial velocities of gas and liquid.

Energy & Fuels

As dH increases from 51 to 72 µm (Samples 4 and 4a), the pressure gradient increases for all the foam qualities tested. We do not have an explanation for this increase in pressure gradient. The bubbles are larger in Sample 4a (Table 6). However, when dH increases further to 207 µm (Sample 4b), the pressure gradient decreases substantially (Fig. 30). dH = 51 µm

1200 Pressure gradient, mbar/m

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 36

dH = 72 µm

dH = 207 µm

1000 800 600 400 200 0

`

0

0.2

0.4

fg

0.6

0.8

1

Figure 30. Samples 4, 4a, and 4b: foam-quality scans at different values of dH. The error bars in the data reflect oscillations in pressure gradient.

As dH increases from 115 to 145 µm (Samples 5 and 5a), the pressure gradient substantially decreases. An additional 17% increase in dH yields only a marginal decrease in pressure gradient (Fig. 31). This appears to be related to the number of bubbles in the two cases. The number of bubbles for dH = 170 µm decreases slightly compared to the case at dH = 145 µm.

ACS Paragon Plus Environment

Page 23 of 36

dH = 115 µm

700 Pressure gradient ,mbar/m

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

dH = 145 µm

dH = 170 µm

600 500 400 300 200 100 0 0

0.2

0.4

0.6

0.8

1

fg Figure 31. Sample 5, 5a, and 5b: foam-quality scans at different dH. The error bars in the data reflect oscillations in pressure gradient.

Table 9 summarizes the pressure-gradient results. For this comparison we selected fg and ut to be in the vicinity of 0.45 and 0.0025 m/s for all the samples. Samples 2 and 5 deviate the most from the selected ut, but based on the shear-thinning behaviour shown in Figs. 24 and 26, we do not expect significant change in the values of ∇Pfoam. Sample 2 showed the highest value of ∇P with foam. We believe this reflects the small dH and Lp (compared to Sample 4, with similar dH). The calculated mobility-reduction factor of foam compared to single-phase flow of water (MRF) is based on the single-phase flow experiments used to determine dH for each sample35. Fig. 32 shows how pressure gradient responds to dh and Lp. Pressure gradient is much greater for narrower fractures. It also increases with decreasing Lp, though less dramatically. There is no simple trend between MRF and either dH or Lp alone (Figs. 33 and 34). Because MRF is a comparison to single-phase laminar flow, if

p decreases with increasing dh less than (dh)(-3), MRF increases. Table 9. Summary of pressure-gradient results with respect to variation in dH and Lp at specific flow conditions.

dH, µm

ut, m/s

fg

MRF

∇P

∇P

(foam),

(water),

mbar/m

mbar/m

Lp, µm

Bubble size, mm2

Sample 1

670

0.0025

0.45

35

0.67

52

2661

NA

Sample 2

66

0.0030

0.38

2466

82.6

30

819

0.089

Sample 3

330

0.0025

0.45

52

2.75

19

5156

4.315

Sample 4

51

0.0032

0.45

713

142.6

5

4415

0.097

Sample 4a

72

0.0032

0.45

800

80

10

4415

0.145

ACS Paragon Plus Environment

Energy & Fuels

Sample 4b

207

0.0032

0.45

137

9.1

15

4415

1.37

Sample 5

115

0.0022

0.45

563

29

19.4

2421

0.468

Sample 5a

145

0.0022

0.45

162

13

12.4

2421

0.74

Sample 5b

170

0.0022

0.45

117

13

9

2421

0.943

Samples 1, 2 and 3

Sample 4, 4a, 4b

6000 713

137 799

4000 Lp, µm

Sample 5, 5a, 5b

52

5000

35

3000 563

2000

116 162

2466

1000 0 0

200

400 dh, µm

600

800

Fig. 32. Effect of pore-geometry parameters dH and Lp on ∇P (mbar/m) with foam (numbers printed next to data points). Lp was fixed in Samples 4. 4a and 4B and in and 5, 5a and 5b.

60.0

Samples 1, 2 and 3 Sample 5, 5a, 5b

Sample 1

6000

50.0 40.0

Sample 2

Sample 5 Sample 3 20.0 Sample 5a Sample 4b Sample 4a 10.0 Sample 5b Sample 4 0.0 0 200 400 dH, µm

Figure 33. MRF versus dH for all samples.

52

3000 19

2000

30

1000 600

800

Fractu re

14.9 10.5

4000

30.0

Sample 4, 4a, 4b 19

5

5000 Lp, µm

MRF

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 36

12.9 12.8

0 0

200

dH400 , µm

600

800

Figure 34. Effect of dH and Lp on MRF (numbers printed next to data points). Lp was fixed in Samples 4, 4a and 4b and in Samples 5, 5a and 5b.

Geom etry and Foam Properties Appendix B presents a characterization of expected foam-generation mechanisms based on fracture geometry, characterized in three dimensionless groups: the ratio of throat width to throat aperture

ACS Paragon Plus Environment

Page 25 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

(wt/dt), the ratio of body aperture to throat aperture (db/dt), and the ratio of body width to body aperture (wb/db). Briefly, one expects repeated snap-off of small bubbles in slit-shaped throats ((wt/dt)  ∞), both at the gas-invasion front and behind the front (due to fluctuating capillary pressure). For throats of width comparable to aperture, snap-off depends on the geometry of the downstream body (its aperture and width). If the body is much deeper than the throat, or much wider than it is deep, snap-off is expected at the gas-invasion front, but resulting bubbles would be larger than for slit-shaped throats. Snap-off behind the front requires larger fluctuations in capillary pressure than for slit-shaped throats. If the body is not much deeper than the throat or much wider than it is deep, snap-off is not favoured, but lamella division is. Table 10 presents a characterization of the model fractures in terms of this analysis. The results agree on the whole with our observations. Snap-off of small bubbles is expected and observed in samples 2 and 4 (Figs. 16 and 18). Conditions are somewhat less favourable for snap-off in Sample 3, and bubbles are larger (Fig. 17), though these bubbles are reduced somewhat in size as they move downstream. Conditions are less favourable for snap-off in samples 4a and 4b than sample 4, and indeed the bubble size increases from sample 4 to 4a to 4b (Fig. 19). Throat and body geometries are not favourable for snap-off in Samples 1 and 5, and the primary mechanism for foam generation appears to be lamella division (Figs. 15 and 20), though bubbles are reduced in size as they propagate in Sample 5. Conditions are less favourable for snap-off in Samples 5a and 5b than in Sample 5, and indeed the bubble size increases from sample 5 to 5a to 5b (Fig. 21).

Table 10. Geometric characterization of model fractures in terms of dimensionless groups

Pore-throat aperture, dt

Pore-body aperture, db

Pore-throat width, wt

(wt/dt)

(db/dt)

Sample 1

818

1128

1550

1.89

1.37

Sample 2

68

138

410

6.0

2

Sample 3

443

853

1650

3.72

1.9

Sample 4

100

210

1130

11.3

2.1

Sample 4a

121

231

1130

9.33

1.91

Sample 4b

253

363

1130

4.46

1.43

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 36

Sample 5

131

211

460

3.51

1.61

Sample 5a

161

241

460

2.86

1.50

Sample 5b

186

265

460

2.47

1.43

Summary and Conclusions Experiments investigating foam generation, propagation and mobility reduction were carried out using a variety of model fractures with different geometries. The following conclusions can be drawn: 1. Foam was generated in situ in different model fractures that varied in the magnitudes of the aperture, aperture variation within the fracture and length scale over which the aperture varies. Foam in the model fractures was generated primarily by two processes: capillary snap-off and lamella division. In both cases the fracture-wall roughness played a major role in foam generation. 2. Two of the five fracture samples show only lamella division. This may reflect relatively wide apertures and a throat geometry less favorable for snap-off (i.e., less slit-like). The other three samples show both generation mechanisms at different foam qualities and superficial velocities. 3. In cases where foam is generated only by lamella division, gas enters the fracture and propagates for some distance as a continuous phase before additional films are created. 4. In all cases, bubbles smaller than the pores are generated and propagate through the fracture. The size of the bubbles is not always similar to the size of the pore, as is thought to be the case in 3D rock pore space, in part because bubbles reside for a time that is much shorter than the time required for diffusion to eliminate small bubbles. Moreover, snap-off can produce bubbles much smaller than pores in slit-shaped throats. 5. Very small pressure gradients were recorded for the samples with very large apertures. In these cases no foam-quality scans could be conducted. In most cases, bubble size increased and pressure gradient declined as the aperture increased for the same roughness of the pore wall. In some cases, however, the mobility reduction factor increased relative to water; that is, as hydraulic aperture increased, the pressure gradient decreased less than the (-3) power of the aperture (as it does for single-phase flow of water).

ACS Paragon Plus Environment

Page 27 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

6. Foam-quality scans were carried out using three samples. The pressure-gradient data reveals, in two of the fractures, high- and low-quality flow regimes like those seen in rock matrix. However, the high-quality regime was controlled not by foam stability and coalescence but by fluctuations in foam generation, and bubble size was not fixed at pore size in the low-quality regime. 7. Hydraulic aperture alone is not enough to determine foam-generation behavior and mobility reduction. The roughness scale, both laterally and vertically, plays a significant role. 8. When the roughness scale was fixed, a significant reduction in pressure gradient was measured with increasing hydraulic aperture. Foam bubbles become larger as the aperture increases.

Acknowledgments The authors acknowledge Saudi Aramco for providing the scholarship for Mr. AlQuaimi, and also the generous support provided by the sponsors of the Joint Industry Project on Foam for Enhanced Oil Recovery at Delft University of Technology. Special thanks go to Sian Jones for her helpful comments and discussion.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

References (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)

(12) (13) (14)

Holm, L., Foam injection test in the Siggins field, Illinois. Journal of Petroleum Technology, 1970. 22(12): p. 1,499-1,506. Svorstøl, I., et al. Foam Pilot Evaluations for the Snorre Field, Part 2: Numerical Simulations and Economical Evaluations. in IOR 1995-8th European Symposium on Improved Oil Recovery. 1995. Vierma, Austria. Svorstøl, I., et al. Foam Pilot Evaluations for the Snorre Field, Part 1: Project and Laboratory Results. in IOR 1995-8th European Symposium on Improved Oil Recovery. 1995. Vierma, Austria. Blaker, T., et al. Foam for gas mobility control in the Snorre field: the FAWAG project. in SPE Annual Technical Conference and Exhibition. 1999. Houston, Texas: Society of Petroleum Engineers. Patzek, T.W., Field applications of steam foam for mobility improvement and profile control. SPE Reservoir Engineering, 1996. 11(02): p. 79-86. Kam, S., et al., Experimental study of high-temperature foam for acid diversion. Journal of Petroleum Science and Engineering, 2007. 58(1): p. 138-160. Nasr-El-Din, H.A., et al. Field Success in Carbonate Acid Diversion, Utilizing Laboratory Data Generated by Parallel Flow Testing. in SPE Annual Technical Conference and Exhibition. 2006. San Antonio, Texas, USA Society of Petroleum Engineers. Szafranski, R., et al., Surfactant/foam process for improved efficiency of aquifer remediation. Structure, Dynamics and Properties of Disperse Colloidal Systems, 1998. 111: p. 162-167. Nelson, R., Geologic analysis of naturally fractured reservoirs. 2001: Gulf Professional Publishing. Ozkaya, S.I. Detection of Fracture Corridors from Openhole Logs in Horizontal Wells. in SPE Saudi Arabia Section Technical Symposium. 2007. Dhahran, Saudi Arabia. SPE-110942-MS: Society of Petroleum Engineers. Bertotti, G., A. Immenhauser, and J.K.J.T.-v. Koppen, Stratigraphic and regional distribution of fractures in Barremian–Aptian carbonate rocks of Eastern Oman: outcrop data and their extrapolation to Interior Oman hydrocarbon reservoirs. International Journal of Earth Sciences, 2005. 94(3): p. 447-461. van Golf-Racht, T.D., Fundamentals of fractured reservoir engineering. 1982: Elsevier. Haugen, Å., et al., Experimental study of foam flow in fractured oil-wet limestone for enhanced oil recovery. SPE Reservoir Evaluation & Engineering, 2012. 15(02): p. 218-228. Haugen, Å., et al., Miscible and immiscible foam injection for mobility control and EOR in fractured oil-wet carbonate rocks. Transport in porous media, 2014. 104(1): p. 109-131.

ACS Paragon Plus Environment

Page 28 of 36

Page 29 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

(15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35)

Steinsbø, M., et al. Foam as mobility control for integrated CO2-EOR in fractured carbonates. in IOR 2015-18th European Symposium on Improved Oil Recovery. 2015. Dresden, Germany 14-16 April 2015: EAGE. Yan, W., C.A. Miller, and G.J. Hirasaki, Foam sweep in fractures for enhanced oil recovery. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2006. 282: p. 348-359. Buchgraber, M., L.M. Castanier, and A.R. Kovscek. Microvisual investigation of foam flow in ideal fractures: role of fracture aperture and surface roughness. in SPE Annual Technical Conference and Exhibition. 2012. Society of Petroleum Engineers. Gauteplass, J., et al., Pore-level foam generation and flow for mobility control in fractured systems. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2015. 468: p. 184-192. Kovscek, A., et al., Foam flow through a transparent rough-walled rock fracture. Journal of Petroleum Science and Engineering, 1995. 13(2): p. 75-86. Ferno, M.A., et al., Experimental Study of Foam Generation, Sweep Efficiency, and Flow in a Fracture Network. SPE Journal, 2016. 21(4): p. 1140-1150. AlQuaimi, B.I. and W.R. Rossen, Study of foam generation and propagation in a fully characterized physical-model fracture. Accepted in the Journal of Petroleum Science and Engineering, 2017. Chen, C.Y., R.N. Horne, and M. Fourar, Experimental study of liquid-gas flow structure effects on relative permeabilities in a fracture. Water Resources Research, 2004. 40(8): p. W08301. Chen, C.Y., K. Li, and R.N. Horne. Experimental study of phase transformation effects on relative permeabilities in fractures. in SPE Annual Technical Conference and Exhibition. 2004. Society of Petroleum Engineers. Fourar, M., S. Bories, and R. Lenormand, Experimental study of two-phase flow in rough fractures. Proceedings, seventeenth workshop on geothermal reservoir engineering. Stanford University, Stanford, California, 1992: p. 215-218. Pieters, D. and R. Graves. Fracture relative permeability: linear or non-linear function of saturation. in International Petroleum Conference and Exhibition of Mexico. 1994. Society of Petroleum Engineers. Qian, J., et al., Experimental study of the effect of roughness and Reynolds number on fluid flow in rough-walled single fractures: a check of local cubic law. Hydrological Processes, 2011. 25(4): p. 614-622. Pruess, K. and Y.W. Tsang, On two-phase relative permeability and capillary pressure of rough-walled rock fractures. Water Resources Research, 1990. 26(9): p. 1915-1926. AlQuaimi, B.I. and W.R. Rossen, Characterizing Foam Flow in Fractures for Enhanced Oil Recovery, in 19th European Symposium on Improved Oil Recovery. 2017: Stavanger, Norway. AlQuaimi, B.I. and W.R. Rossen, New Capillary Number Definition for Displacement of Residual Nonwetting Phase in Natural Fractures. Geophysical Research Letters, 2017. Rossen, W. and A.T. Kumar, Single-and two-phase flow in natural fractures, in SPE Annual Technical Conference and Exhibition. 1992, Society of Petroleum Engineers: Washington, D.C., USA. Tsang, Y., The effect of tortuosity on fluid flow through a single fracture. Water Resources Research, 1984. 20(9): p. 1209-1215. Hughes, R.G. and M.J. Blunt, Network modeling of multiphase flow in fractures. Advances in Water Resources, 2001. 24(3–4): p. 409-421. Pyrak-Nolte, L.J., N.G. Cook, and D.D. Nolte, Fluid percolation through single fractures. Geophysical Research Letters, 1988. 15(11): p. 1247-1250. AlQuaimi, B.I. and W.R. Rossen, Capillary desaturation curve additional figures. 2017: Delft University of Technology Reposotiry http://dx.doi.org/10.4121/uuid:b959da2e-f955-4991b2bc-b2514397671f. AlQuaimi, B.I. and W.R. Rossen, Capillary Desaturation Curve for Residual Nonwetting Phase in Natural Fractures. SPEJ., 2017.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(36) (37) (38) (39) (40) (41)

Rossen, W.R., Foams in enhanced oil recovery in R. K. Prud'homme and S. Khan, ed., Foams: Theory, Measurements and Applications. 1996, New York: Marcel Dekker. 413-464. Rossen, W.R., A critical review of Roof snap-off as a mechanism of steady-state foam generation in homogeneous porous media. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2003. 225(1): p. 1-24. Ransohoff, T. and C. Radke, Mechanisms of foam generation in glass-bead packs. SPE reservoir engineering, 1988. 3(02): p. 573-585. Khatib, Z., G. Hirasaki, and A. Falls, Effects of capillary pressure on coalescence and phase mobilities in foams flowing through porous media. SPE reservoir engineering, 1988. 3(03): p. 919-926. Alvarez, J.M., H.J. Rivas, and W.R. Rossen, Unified Model for Steady-State Foam Behavior at High and Low Foam Qualities. SPE journal, 2001. Osterloh, W. and M. Jante. Effects of gas and liquid velocity on steady-state foam flow at high temperature. in SPE/DOE Enhanced Oil Recovery Symposium. 1992. Society of Petroleum Engineers.

(42)

Lenormand, R., Zarcone, C., and A. Sarr, A. Mechanisms of the displacement of one fluid by another in a network of capillary ducts. J. Fluid Mech, 1983. 135, 337.

(43)

Rossen, W.R. and Gauglitz, P.A. Percolation theory of creation and mobilization of foams in porous media. AIChE J., 1990. 36, 1176-1188.

(44)

Rossen, W.R. Theory of mobilization pressure gradient of glowing goams in porous media. I. incompressible foam. J. Colloid Interface Sci., 1990, 136, 1-16.

(45)

AlQuaimi, B. I. Investigation of foam generation, propagation and rheology in fractures. (Doctoral dissertation, Delft University of Technology), (2017).

ACS Paragon Plus Environment

Page 30 of 36

Page 31 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Appendix A. Topographic Maps and 2D Pore Networks of fracture samples Figs. A-1 to A-5 present topographic maps of the roughened glass surfaces from which our

fracture models are made. Superimposed over the images are the conceptual 2D networks implied by that topography if wetting phase occupies locations of narrow aperture and nonwetting phase locations of wider aperture. Pore bodies are outlined in red, and throats marked with red dots. Figures are taken from Ref. (21).

Figure A-1. Sample 1: 2D network superimposed on 3D surface topography of the fracture.

Figure A-2. Sample 2: 2D network superimposed on 3D surface topography of the fracture.

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure A-3. Sample 3: 2D network superimposed on 3D surface topography of the fracture.

Figure A-5 Sample 5: 2D network superimposed on 3D surface topography of the fracture.

Figure A-4. Sample 4: 2D network superimposed on 3D surface topography of the fracture.

ACS Paragon Plus Environment

Page 32 of 36

Page 33 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Appendix B: Pore Geometry and Mechanisms of Lamella Creation Snap-Off For gas to penetrate a pore throat with a circular cross-section, the capillary pressure Pc must exceed the capillary entry pressure of the throat Pce: Pc > Pce = 2/(Rt cos),

(B1)

where  is surface tension, Rt the radius of the throat, and  contact angle of the gas-water interface. For simplicity, in the remainder of this discussion we assume perfect wetting ( = 0). The requirement derives from the hemispherical shape of the gas-water interface as it penetrates the throat. Once gas fills the throat, if the throat is smoothly constricted (i.e., nearly cylindrical along its length), then there is a cylindrical interface of radius Rt at the wall. If capillary pressure falls below (/Rt), the film swells and snap-off results. Thus, if, after gas first penetrates the throat, local capillary pressure falls to less than Pce/2, and snap-off occurs in the throat. If the downstream pore body has a radius Rb, then, as gas first penetrates the body, the curvature of the spherical gas-water interface is (2/Rb). Thus, if Rb > (2 Rt), there is a moment where local capillary pressure is low enough for snap-off in the upstream pore throat. This mechanism (called "Roof snap-off") stops when the downstream pore body fills with gas, but there are other possible origins of fluctuations in local capillary pressure that can trigger snap-off37. For a pore throat with rectangular cross section, of dimensions wt x dt, (with width wt > dt) the capillary condition for gas penetration of the throat is approximately42 Pc > Pce = [/(wt/2) + /(dt/2)]

(B2)

Thus, for a throat with square cross section, Pce is roughly that for the cylindrical throat with the same width37. If wt >> dt, the interface is nearly cylindrical, not spherical, as gas penetrates the throat, with Pce  [/(dt/2)]. After gas invasion, snap-off occurs when the liquid behind the cylindrical interfaces on two sides of the throat swell and the interfaces approach and meet; see Fig. B1. This occurs if Pc < [/(dt/2)]. Thus, for a slit-shaped throat, a slight, temporary decrease in capillary pressure from Pce can trigger snap-off. During gas invasion of the downstream pore body, this occurs as soon as the interface passes beyond the throat and reaches a sufficient increase in aperture d. The bubble formed moves into the pore body, the throat fills again with liquid, and the process can repeat.

ACS Paragon Plus Environment

Page 34 of 36

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(dt/2)

dt wt

Figure B1. Liquid (grey) in a rectangular pore throat on the verge of snap-off.

For a square throat, snap-off during gas invasion depends on the dimensions of the pore body, wb and db. The capillary pressure across the gas-water interface in the pore body is given by Eq. B2, with wt and dt replaced by wb and db. For snap-off, capillary pressure must fall by roughly half, as for a cylindrical throat; i.e. Pc = [/(wb/2) + [/(db/2)] < [/(dt/2)]

(B3)

If the pore body is about as wide as it is deep (wb ~ db), snap-off occurs when the interface is deep enough to satisfy Eq. B3. If the pore body is much wider but not much deeper than the throat (wb >> db ~ dt)), then the term with wb in Eq. A3 is not significant; snap-off occurs at a position wide enough that the first term in brackets in Eq. B3 is insignificant. This is likely to lead to relatively large bubbles. The criteria for snap-off are complicated somewhat by details of pore and throat geometry; see Ref. (37). Snap-off behind the gas-invasion front depends on fluctuations in capillary pressure relative to the prevailing capillary pressure37. T In addition, once bubbles are created, there is a lower limit to the size at which bubbles can divide. For pores of circular cross-section, bubbles of radius less than 2Rt cannot be created by snap-off, because the curvature of the bubble itself imposes a capillary pressure on the bubble. For slit-shaped throats there is again a lower limit to bubble size, which depends on the curvature of the disk-shaped bubbles in the slit geometry.

Lamella Division Lamella division occurs when a moving lamella passes through a gas-filled pore and in the process passes by a pore throat not occupied by liquid or by a lamella36,38. The lamella divides in two and deposits one of the resulting lamellae in the unoccupied pore throat. If snap-off has already filled the gas-occupied porespace with small bubbles, there are no unoccupied throats, and division does not occur. Lamella division requires that bubbles be first displaced out of pore throats43. The pressure difference required to displace such a lamella in a long column of gas depends on details of throat geometry, and in particular on the angle of the pore wall leading from the throat to the body44. If

ACS Paragon Plus Environment

Page 35 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

pore bodies are not much deeper than pore throats, mobilization is easier and lamella division is enhanced. Thus the same conditions that make snap-off less likely ((db/dt) not much greater than 1) tend to favour lamella division.

Dimensionless Groups These findings suggest preliminary dimensionless criteria for snap-off and lamella division in fractures: For slit-shaped throats ((wt/dt)  ∞), one expects repeated snap-off of small bubbles at the leading edge of the gas front. Also, one expects large bubbles to be broken into smaller bubbles behind the foam front by snap-off due to small fluctuations in capillary pressure. For square pore throats ((wt/dt) ~ 1) with square pore bodies ((wb/db) ~ 1), snap-off requires deep pore bodies (db/dt) > 2; snap-off of relatively larger bubbles is expected at the leading edge of the foam front. If the pore body is not much deeper than the throat ((db/dt) ~ 1) but it is very wide ((wb/db)  ∞), snap-off of large bubbles would expected as gas fills the pore bodies. In either case, the fluctuations in capillary pressure required to trigger snap-off behind the drainage front would be greater. Lamella division is not expected to be important if snap-off already has filled the porespace with bubbles of pore size or smaller. If conditions are not good for snap-off and (db/dt) is not much greater than 1, then lamella division could be expected at relatively low pressure gradient. Once foam generation begins, of course, p rises and conditions for lamella division are enhanced. These criteria are summarized in Table B1. These are, of course, only preliminary conclusions and must be confirmed by further experimental studies. One implication of these results is that for samples 4, 4a and 4b, and for 5, 5a and 5b, with the same roughness pattern on the bottom plate but different distances to the top plate, (wt/dt) increases with increasing distance to the top plate, and (db/dt) decrease. One expects less snap-off, and snap-off of larger bubbles, as the distance between the plates increases. Meanwhile, the condition for lamella mobilization and division becomes less difficult.

Table B1. Geometric criteria for snap-off and lamella division.

(db/dt) ~1 and (wb/dt) ~ 1 (wt/dt) ~ 1

(db/dt) > 2 or (wt/dt)  ∞

Relatively little snap-off at foam Snap-off of relatively large bubbles at

ACS Paragon Plus Environment

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

front. Conditions favourable for foam front. Perhaps some snap-off lamella division behind foam front. (wt/dt)  ∞

behind foam front from fluctuating Pc.

Repeated snap-off of small bubbles at foam front. Continued snap-off of larger bubbles into smaller bubbles behind foam front from fluctuating Pc.

Other factors in foam generation Injected gas fraction affects foam (re-)generation behind the front. If little water is injected with the gas to replace water that is displaced, foam generation is of course harder43, and coalescence more likely39. Because capillary pressure controls snap-off, increasing gas fraction in particular inhibits snap-off behind the gas front. Moreover, gas and liquid segregated within the injection trough of our apparatus. This meant that injected fractional flow could not be held constant in some of our experiments. Liquid tended to enter until the liquid level in the trough fell below the level of the fracture, and then gas until the liquid level rose again. This could have been a contributing cause of fluctuating foam generation in our experiments at high foam quality. Greater injection velocity enhances lamella creation by lamella division by increasing the local pressure gradient available to displace liquid lenses and lamellae from pore throats43 and initiate lamella division. All these factors require further study.

ACS Paragon Plus Environment

Page 36 of 36